Quantitative Identification of Metastable Magnesium CarbonateMinerals by Solid-State 13C NMR SpectroscopyJeremy K. Moore,† J. Andrew Surface,† Allison Brenner,† Philip Skemer,§ Mark S. Conradi,‡,†

and Sophia E. Hayes*,†

†Department of Chemistry, Washington University, One Brookings Drive, Saint Louis, Missouri, 63130, United States‡Department of Physics, Washington University, One Brookings Drive, Saint Louis, Missouri, 63130, United States§Department of Earth and Planetary Sciences, Washington University, One Brookings Drive, Saint Louis, Missouri, 63130, UnitedStates

*S Supporting Information

ABSTRACT: In the conversion of CO2 to mineral carbonatesfor the permanent geosequestration of CO2, there are multiplemagnesium carbonate phases that are potential reactionproducts. Solid-state 13C NMR is demonstrated as an effectivetool for distinguishing magnesium carbonate phases andquantitatively characterizing magnesium carbonate mixtures.Several of these mineral phases include magnesite, hydro-magnesite, dypingite, and nesquehonite, which differ incomposition by the number of waters of hydration or thenumber of crystallographic hydroxyl groups. These carbonatesoften form in mixtures with nearly overlapping 13C NMRresonances which makes their identification and analysisdifficult. In this study, these phases have been investigatedwith solid-state 13C NMR spectroscopy, including both static and magic-angle spinning (MAS) experiments. Static spectra yieldchemical shift anisotropy (CSA) lineshapes that are indicative of the site-symmetry variations of the carbon environments. MASspectra yield isotropic chemical shifts for each crystallographically inequivalent carbon and spin−lattice relaxation times, T1, yieldcharacteristic information that assist in species discrimination. These detailed parameters, and the combination of static and MASanalyses, can aid investigations of mixed carbonates by 13C NMR.

■ INTRODUCTION

Many CO2 capture and sequestration studies have focused onthe formation of magnesium carbonate minerals from mafic andultramafic protoliths as a permanent storage solution for CO2.These reaction products are often imprecisely described asmagnesite (MgCO3), the energetically favored magnesiumcarbonate. There are, however, additional thermodynamicallymetastable phases that include waters of crystallization andhydroxyl groups in the crystal structure.1,2 These metastableminerals form due to kinetic trapping along the reactionpathway to magnesite. The lifetime of the metastablemagnesium carbonate minerals can be long, even relative togeologic time scales, since the energy needed to remove a waterfrom the crystal structure is not always available to the system.1

There are many possible forms of these hydroxy-hydratedmagnesium carbonates: some examples, listed from leasthydrated to most, include hydromagnesite [4MgCO3·Mg-(OH)2·4H2O], dypingite [4MgCO3·Mg(OH)2·(5−8)H2O],and nesquehonite [Mg(OH)(HCO3)·2H2O].In CO2 sequestration reactions, it has been reported that Mg-

containing minerals react with CO2 to form products consistingof a mixture of hydroxy-hydrated magnesium carbonates and

magnesite.1,3−7 The specific composition of these productmixtures is dependent on the reactant minerals and reactionconditions, such as temperature and pH.1,8−11

The ability to distinguish and quantify the product mineralsin a mixture of magnesium carbonate phases is difficult due to acombination of factors. One complicating factor is that reactionproducts often form mixtures of amorphous and crystallinespecies,12−14 and current methods such as powder X-raydiffraction (pXRD) cannot identify the amorphous compo-nents. Additionally, multiple possible hydrated magnesiumcarbonates can be produced as well as adducts that form withions, such as Na+ and Ca2+. Here we will demonstrate theability of NMR to identify these mineral phases andquantitatively characterize these species in mixtures, therebyfacilitating the study of sequestration reactions and theirkinetics.

Received: July 14, 2014Revised: October 28, 2014Accepted: December 1, 2014

Article

pubs.acs.org/est

© XXXX American Chemical Society A dx.doi.org/10.1021/es503390d | Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Solid-state nuclear magnetic resonance (NMR) spectroscopyis a powerful tool for identification of various carbonatespecies,2,14−19 despite having relatively small structural differ-ences between them. Pure MgCO3 phases are needed asstandards for rigorous NMR characterization, and we presentsynthesis methods to obtain these for each of the metastableminerals. The NMR signatures of magnesite, hydromagnesite,dypingite, and nesquehonite are established through a suite of13C NMR techniques from the pure mineral samples, withassignments confirmed by Raman spectroscopy and pXRD.NMR measurements used for this work include 13C{1H}

magic angle spinning (MAS) NMR with proton decoupling(denoted by 1H in brackets) and 13C{1H} cross-polarizationmagic angle spinning (CPMAS) NMR. When employed, MASNMR narrows resonances that are broadened by dipolarcoupling, as well as chemical shift anisotropy (CSA). As aresult, narrow resonances are observed at a species’ isotropicchemical shift (δiso). MAS NMR allows the minerals to bedistinguished by both the chemical shift and the number ofpeaks which corresponds to the number of crystallographicallyinequivalent 13C sites. Static 13C NMR yields a characteristicline shape from broadening of the NMR resonance through theCSA. These broad “powder patterns” can be used to distinguishthe symmetry of the chemical environment of the 13C in theproduct mineral.15,20−23 Static 13C NMR resonances aredefined by components evident in the CSA line shapenamely, their isotropic chemical shift (δiso), chemical shiftanisotropy (δaniso), and asymmetry (η). These parameters aredetermined by the shape of the static NMR CSA powderpattern and are unique for each metastable magnesiumcarbonate mineral based on the carbon atoms’ nearbyenvironments.With this detailed characterization, it is possible to

distinguish these products in a variety of settings, includinganalysis from in situ and ex situ NMR experiments.2,19 The 13CMAS NMR, pXRD, and Raman spectroscopy each lendthemselves to ex situ characterization of subsets of the productminerals which allow, for example, the study of compositionalgradients within a reactor. The 13C static NMR gives thecapability for in situ monitoring of the mineral and CO2reaction ensemble that forms these mixtures of metastablemagnesium carbonates. Using the magnesium carbonate phasesand the quantitative characterization afforded by solid-stateNMR, the capability of distinguishing and quantifying kineti-cally trapped magnesium carbonate minerals is possible in bothin situ and ex situ NMR experiments.

■ MATERIALS AND METHODSMagnesite, MgCO3, was synthesized in a high temperature,elevated pressure vessel for use in analyzing the reaction in situwith 13C static NMR as described previously.2 Samples wereprepared by adding 1.60 g of forsterite, Mg2SiO4, powder and2.77 mL deionized (DI) water into the reaction vessel. Thevessel was then pressurized with 110 atm of 13CO2 generatedthrough cryogenic pressure intensification and then heated to90 °C. The mixture was then allowed to react for 8 days.Magnesite is the thermodynamically favored product fromreactions with forsterite at these elevated temperatures andpressures, and therefore was formed preferentially.1,3,24−27 Themagnesite precipitates on the surface of the forsterite particles.Hydromagnesite, [4MgCO3·Mg(OH)2·4H2O], was synthe-

sized28,29 by adding 0.107 g of 99% 13C enriched NaHCO3 to asolution of 0.245 g of MgCl2 in 8.0 mL of DI water. The

solution was mixed for 1 min, and then allowed to react for 4days at 92 °C. The precipitate was filtered and dried undervacuum for 24 h at room temperature.Dypingite, [4MgCO3·Mg(OH)2·(5−8)H2O], was synthe-

sized7,28,29 by adding 0.221 g of 99% 13C enriched NaHCO3to a solution of 0.251 g of MgCl2 in 3.25 mL of DI water. Thissolution reacted for 24 h at room temperature, followed by 48 hof reaction at 62 °C, finally reacting for 24 h at 72 °C. Theprecipitate was then filtered and dried under vacuum for 24 h atroom temperature.Nesquehonite, [Mg(OH)(HCO3)·2H2O], was synthe-

sized28−30 by adding 0.146 g of 99% 13C enriched NaHCO3to a solution of 0.164 g of MgCl2 in 3.25 mL of deionized (DI)water.30 The solution was stirred for 1 min, and the glass beakerwas sealed. The solution reacted for 3 days at roomtemperature. The precipitate was collected by filtering thesample, drying under vacuum at 40 °C for 10 min, followed byair drying overnight. The powder was then rinsed with DI waterbefore repeating the filtering and drying procedure. Themineral assignments were confirmed by pXRD and 13C NMR.The NMR measurements were acquired on a 7 T magnet

with a 1H frequency of 299.673 MHz and a 13C frequency of75.365 MHz with a commercial HX MAS Chemagnetics probe.13C{1H} MAS and static NMR spectra were acquired in a singlescan by a Bloch decay sequence with decoupling duringacquisition (28.4 kHz typical decoupling strengths) spinningfrequency (νR, MAS only) of 5 kHz, using a π/2 pulse length of8.8 μs, and allowing the sample to reach thermal equilibrium inthe magnetic field over 45 min. 13C{1H} CPMAS cross-polarization “build-up” curves and spectra were acquired withνR of 5 kHz, a recycle delay of 5 s, and recording 32 transients.The B1 field strengths were 27.8 kHz which gave a 1H π/2pulse length of 9 μs. The matching condition for the contactperiod was found by the Hartmann−Hahn match and the B1fields were not modulated during the contact time. Typicalcross-polarization contact times (τcp) of 100 μs to 15 ms wereused to obtain the “build-up” curves. Chemical shifts werereferenced to adamantane at 29.45 ppm. The isotropic chemicalshifts for the static spectra were assigned via peak fitting andconfirmed with slow spinning MAS NMR. Some samples werecomixed with a carbon-free powder, forsterite, due to limitedsample quantity (for the purpose of filling a NMR rotorvolume).The relaxation time T1 was measured by 13C{1H} MAS

NMR saturation recovery pulse sequence and fitting the peakarea versus delay time curve. The T1ρ and TIS were measuredwith 13C{1H} CPMAS NMR and fitting the cross-polarization“build-up” curve. The peak areas were collected from the fits ofthe data.All NMR spectra were fit with DMFIT.31 δiso is the isotropic

chemical shift and is defined as 1/3(δXX+δYY+δZZ). Theprincipal values of the CSA powder pattern are defined as|δZZ−δiso|>|δXX−δiso|>|δYY−δiso|. The anisotropic chemical shiftis defined as δaniso = δZZ−δiso. Finally, the asymmetry is definedas η = (δYY−δXX)/δaniso.The Raman spectra were acquired on a HoloLab Series 5000

Laser Raman Microprobe with a 532 nm laser beam at 11 mWof power. The resolution was approximately 3 cm−1, andsampled regions were approximately 5 μm in diameter. Thereproducibility of the Raman data were better than 1 cm−1.The pXRD data were obtained on a Rigaku Geigerflex D-

MAX/A diffractometer with Cu−Kα radiation at 35 kV and 35mA. Values of 2θ ranged from 6° to 60° with a step size of

Environmental Science & Technology Article

dx.doi.org/10.1021/es503390d | Environ. Sci. Technol. XXXX, XXX, XXX−XXXB

0.04° and 1 s intervals. Samples were held in a standard pXRDslide. Samples that had been comixed with forsterite, for theNMR studies, have peaks for this mineral in the pXRD patterns(as seen in Supporting Information (SI) Figures S2 and S3).

■ RESULTS AND DISCUSSIONThe NMR spectrum of each mineral, both static and MAS, willbe discussed individually, while the nuclear relaxation data willbe discussed at the end, as to easily compare the minerals. Allthe relaxation data and CSA parameters are summarized inTable 1 and the CSA parameters are defined in the Materialsand Methods section. The assignment of MAS NMR peaks wasconfirmed by slow spinning MAS NMR, which is discussed inthe SI and shown in Figure S1. Schematics of the carbon sitespresent in the metastable magnesium carbonates are shown inScheme 1. Shown are an axially symmetric CO3

2− carbonate

(D3h point group) and a lower symmetry HCO3− bicarbonate

species. Distinguishing between CO32− and HCO3

− is possiblevia their 13C δiso chemical shifts, which have shifts of 170 and160 ppm, respectively, in solution. Alternatively, the CSA lineshape can be unambiguous for characterization in static 13CNMR.Magnesite. Magnesite, MgCO3, precipitates during seques-

tration reactions on the surface of forsterite, the reactantmineral phase.32,33 It forms directly in solutions with Mg2+ andCO3

2− ions at temperatures above 120 °C but will form via anintermediate, hydromagnesite (that will be discussed later),even when the temperature is under 120 °C if given sufficienttime. Magnesite is the only nonhydrated magnesium carbonateand has a single 13C NMR resonance at 169.9 ppm as seen instatic 13C{1H} NMR (Figure 1A) and 13C{1H} MAS NMR(Figure 1C, 1D).The static spectrum has an asymmetry (η) of 0.14 and has a

δaniso of −54.5 ppm. The asymmetry of pure crystallinemagnesite is predicted to be 0, by the D3h crystal symmetryof the 13C carbonate site (as seen in Scheme 1), suggesting thatsites within magnesite have disorder from the surfaces or grainboundaries between the magnesite particles and forsterite

particles which cause deviations from pure MgCO3. It has beenreported previously that calcite (CaCO3) can precipitate withprotons in the crystal,14,20 which could also account for anonzero asymmetry of the powder pattern. Further evidence ofprotons in magnesite’s crystal structure can be seen in aseparate reaction that ran for a shorter time as seen in Figure 2,where the magnesite peak appears with a shoulder.This pair of resonances, seen in Figure 2, represents a

magnesite phase with nearby protons and a proton-freemagnesite phase. Both samples from Figure 1and Figure 2appear as magnesite alone in pXRD and Raman spectroscopy

Table 1. Summary of NMR Parameters and Relaxation Time Constants

minerals

magnesite hydromagnesite dypingite nesquehonite13C δiso (ppm) 169.9 165.2 163.1 162.5 165.3 163.5 165.4

% area for each C site 100 51.8 48.4 60.7 39.3 10013C δxx (ppm) 204.2 193.9 203.9 193.0 205.8 204.113C δyy (ppm) 196.8 186.4 175.0 184.9 175.1 178.113C δzz (ppm) 118.8 115.8 110.5 118.0 109.6 115.113C η 0.14 0.16 0.55 0.17 0.57 0.5113C δaniso (ppm) −54.5 −47.4 −49.1 −47.3 −53.9 −50.613C T1 (s) 78.0 157 156 104 107 35.71H T1ρ (ms) 7.1 3.5 8.0 4.8 3.7 58

TIS (ms) 0.59 0.75 0.38 0.49 0.60 1.1

Scheme 1. Schematic of Carbon Site Symmetry for CO32−

(Left) and HCO3− (Right)

Figure 1. 13C NMR of a magnesite (MgCO3) mineral species grownon the surface of forsterite (Mg2SiO4). (A) Static

13C{1H} NMR. (B)Fit of the static 13C NMR with parameters listed in Table 1. (C)13C{1H} MAS NMR with νR = 5 kHz. (D) Expanded view of 13C{1H}MAS NMR. * denotes spinning sideband.

Environmental Science & Technology Article

dx.doi.org/10.1021/es503390d | Environ. Sci. Technol. XXXX, XXX, XXX−XXXC

(seen in SI Figure S2). It is also possible to cross-polarize theproton-containing magnesite peak with CPMAS proving thatthere are indeed protons nearby the magnesite carbon (CO3

2−)in the higher frequency peak. The shift to higher frequency wasreported in a hydrated calcium carbonate.14 Other evidenceused in assigning the higher frequency peak to the proton-containing magnesite species is comparing the peak fwhm. Thefwhm of the peak in Figure 1 is 0.46 ppm. The peak at 169.5ppm in Figure 2 has a fwhm of 0.57 ppm while the fwhm of thepeak at 170.2 ppm is 1.08 ppm. Since the low frequency peak inFigure 2 and the peak in Figure 1 have a similar fwhm, thestructures have similar crystallinity which can be dictated by thepresence of protons. The high frequency peak in Figure 2,however, is broadened, presumably by nearby protons.Hydromagnesite. Hydromagnesite forms in sequestration

reactions as a mixture with dypingite, described below, insolutions with high concentrations of Mg2+, HCO3

−, andCO3

2− and temperatures between 40 and 120 °C.1,6,28

However, hydromagnesite can be synthesized in isolated formin benchtop reactions. There are two crystallographicallyinequivalent carbons in the crystal structure of hydro-magnesite,34−36 seen in the 13C NMR spectra of Figure 3,and the assignment has been confirmed with pXRD (SI FigureS3). Hydromagnesite is the least hydrated metastablemagnesium carbonate, the H2O/Mg ratio being 0.8, thecomposition of hydromagnesite being [4MgCO3·Mg(OH)2·4H2O].The two resonances have isotropic 13C chemical shifts of

165.2 and 163.0 ppm. These peaks, respectively, account for51.8% and 48.2% of the integrated area signifying that the twosites are approximately equally populated in the mineral, asexpected. The asymmetry of the CSA line shape shows onepeak (δiso = 165.2 ppm) as nearly axially symmetric and theother (δiso = 163.0 ppm) as more asymmetric with η values of0.16 and 0.55 respectively. These CSA parameters indicate thatone of the carbon sites in the crystal structure is similar to thatof magnesite, which is axially symmetric, while the other sitehas an asymmetry similar to the asymmetric nesquehonitecarbon, as will be seen below.When using CPMAS NMR, the resonance at 163.0 ppm split

into two resonances (as seen in SI Figure S4) with differentvalues for the time constants T1ρ and TIS for each resonance.This split resonance reveals why the peak at 163.0 ppm isshorter and broader than the peak at 165.2 ppm: the peak atlower frequency appears to be made up of at least tworesonances. These data indicate the nesquehonite-like carbon,163.0 ppm, has more disorder caused by multiple config-urations, likely due to nearby waters of crystallization.

Dypingite. Dypingite forms in sequestration reactions as amixture with hydromagnesite in solutions with high concen-trations of Mg2+, HCO3

−, and CO32− and temperatures

between 40 and 120 °C.7,10,37,38 Here we were able to isolatea nearly pure dypingite mineral. There is no known crystalstructure for dypingite, but the mineral has a unique pXRDpattern (SI Figure S5).7,38−40 The current convention is thatdypingite is any hydroxy-hydrated magnesium carbonate with 5to 8 waters of hydration7,11 and the formula [4MgCO3·Mg(OH)2·xH2O], (where x = 5−8) giving a H2O/Mg ratio of1−1.6. This makes a pure phase difficult to synthesize;nevertheless, this mineral still has distinguishable NMRfeatures.19 Figure 4 shows the static and MAS 13C{1H} NMRof the dypingite sample.Two NMR resonances are apparent in the 13C MAS NMR

spectrum (see Figure 4C and 4D) with isotropic chemical shiftsof 165.3−163.5 ppm, and each has a unique CSA line shape.There is a small impurity resonance (Figure 4D) at 166.5 ppm(which notably does not match the δiso chemical shift of anyother species we’ve observed so far). These resonances indicatethe presence of two crystallographically inequivalent carbons.The difference is apparent in the asymmetry of the powderpatterns, where the peak at δiso = 165.3 ppm is magnesite-likewith an η value of 0.17 and a δaniso value of −47.3 ppm, while

Figure 2. 13C{1H} MAS NMR of a mineral containing magnesite andmagnesite with nearby protons with νR = 5 kHz.

Figure 3. 13C NMR of a hydromagnesite, [4MgCO3·Mg(OH)2·4H2O], mineral. (A) Static 13C{1H} NMR. (B) Fit of the static 13CNMR with the overall fit in black and the individual sites fitted in redand green and parameters listed in Table 1. (C) 13C{1H} MAS NMRwith νR = 5 kHz. The peak at 165.2 ppm is the green powder patternand the peak at 163.0 ppm is the red powder pattern. (D) Expandedview of 13C{1H} MAS NMR.

Environmental Science & Technology Article

dx.doi.org/10.1021/es503390d | Environ. Sci. Technol. XXXX, XXX, XXX−XXXD

the peak at δiso = 163.5 ppm is more asymmetric with an ηvalue of 0.57 and a δaniso value of −53.9 ppm. Even without aknown crystal structure, the appearance of two crystallo-graphically inequivalent carbon sites in dypingite is encouragingbecause the composition only differs from hydromagnesite(with two unique carbons) by the number of waters ofhydration. The resonance at 165.3 ppm comprises 60.7% of theintegrated area of the spectrum, meaning that approximately 3out of every 5 carbon sites are in the magnesite-like electronicenvironment. To identify dypingite with 13C NMR it isimportant to check the area ratio of the peaks since this arearatio of 3:2 for the resonances at 165.3 and 163.5 ppm isdifferent from that of hydromagnesite (with its 1:1 ratio for thetwo sites), since the two chemical shift values are very close.Nesquehonite. The mineral nesquehonite is the most

hydrated magnesium carbonate phase included here, with aH2O/Mg ratio of 2, and has a chemical composition of[Mg(OH)(HCO3)·2H2O].41 This species has also beendescribed as[MgCO3·3H2O],

30,42,43 which will be discussedmore below. This hydroxy-hydrated magnesium carbonate isthe preferential product in systems with high concentrations ofMg2+, HCO3

−, and CO32− and temperatures below 40

°C.28,30,44,45 The crystal structure of nesquehonite containsonly one crystallographically inequivalent carbon.43,46 There-

fore, only one NMR resonance is expected, and this resonancecan be distinguished in both the static (Figure 5A) and MAS13C{1H} NMR (Figure 5C, and D) spectra and is confirmed bypXRD and Raman in SI Figure S6.The isotropic chemical shift, which can be seen in Figure 5C

and 5D, is 165.4 ppm. The fit of the static spectrum gives anasymmetry (η) of 0.51, a symmetry close to that of thebicarbonate structure seen in Scheme 1, and an anisotropic shift(δaniso) of −50.64 ppm. We note that some previous studiesreport that nesquehonite has two 13C resonances,26,47 but thecrystal structure from X-ray diffraction dictates otherwise. Fromfurther studies (as seen in SI Figure S7), we conclude that thissecond resonance is a northupite impurity in nesquehonite,arising from incomplete sample purification and the presence ofexcess Na+ from the synthesis. This assignment is confirmed bypXRD (SI Figure S8B−E), Raman (SI Figure S8A), and23Na{1H} MAS NMR (SI Figure S9).The large degree of asymmetry of the static line shape

suggests that the symmetry of the 13C site is far from that of D3h(see Scheme 1). The chemical shift is at a lower frequency thanmagnesite suggesting that the complex [Mg(OH)(HCO3)·2H2O] (as opposed to MgCO3·3H2O) is the more accuratedescription due to its variation from the carbonate-likeenvironment. Confirming evidence based on the relaxationtimes is discussed below.

Comparison of Minerals and Techniques. Nuclear spinrelaxation times (13C T1,

1H T1ρ, and TIS) can be used todetermine the carbon atom environment and the dynamics ofnearby protons in the mineral species. A summary of the

Figure 4. 13C NMR of a nearly pure dypingite, [4MgCO3·Mg(OH)2·(5−8)H2O], mineral. (A) Static

13C{1H} NMR. (B) Fit of the static13C NMR with the overall fit in black and the individual sites fitted inred and green and parameters listed in Table 1. (C) 13C{1H} MASNMR with νR = 5 kHz. The peak at 165.3 ppm is the green powderpattern and the peak at 163.5 ppm is the red powder pattern. (D)Expanded view of 13C{1H} MAS NMR. * denotes spinning sideband.

Figure 5. 13C NMR of a nesquehonite mineral. (A) Static 13C{1H}NMR. (B) Fit of the static 13C NMR with parameters listed in Table 1.(C) 13C{1H} MAS NMR with νR = 5 kHz. (D) Expanded view of13C{1H} MAS NMR. * denotes spinning sideband.

Environmental Science & Technology Article

dx.doi.org/10.1021/es503390d | Environ. Sci. Technol. XXXX, XXX, XXX−XXXE

nuclear spin time constants is presented in Table 1 in additionto the principal values of the 13C CSA tensor and CSAparameters previously discussed. The isotropic chemical shiftswere determined from the 13C{1H} MAS NMR spectra.The 13C T1 (spin−lattice) relaxation times across all four

minerals reveal a trend that may prove predictive for othermineral phases. The T1 time increases from nesquehonite todypingite to hydromagnesite. This trend indicates that themore hydrated the mineral phase, the faster the T1 relaxation,due to coupling with the H2O in the crystal structure. The T1for magnesite is shorter than the hydroxy-hydrated carbonatesdespite the expectation of a longer T1 relaxation time in thisnonhydrated mineral. Possible explanations are that para-magnetic impurities dominate the T1 relaxation in magnesite orthat the proximity of the 13C enriched carbonates in magnesitewill enhance the effects of the homonuclear dipole−dipolespin−lattice relaxation mechanism.48 (Note: the homonucleardipole−dipole T1 mechanism is not present in naturalabundance magnesite and therefore, is expected to have alonger T1.) The relaxation time trend is important informationwhen studying new mineral phases since the T1 analysis ensuresquantitative characterization in both in situ and ex situ CO2mineralization reactions and is informative of the level ofhydration of the mineral phase.The 1H T1ρ relaxation time is in the rotating frame, and TIS is

the CP-buildup time governed by 13C−1H dipolar coupling. Innesquehonite, the relatively long T1ρ time constant couldindicate that the protons near the carbon in this mineral havelittle motion. As stated before, this structure has been publishedas both [MgCO3·3H2O] and [Mg(OH)(HCO3)·2H2O]. Forthe nearest proton to be relatively static, the latter of these, themagnesium bicarbonate structure, is the one most consistentwith the T1ρ data (and is consistent with the 13C line shapeshown in Figure 5A). The shorter T1ρ in dypingite andhydromagnesite, may reflect nearby proton dynamics from thewaters of crystallization; however, it is also possible that thepurity and crystallinity of the sample could lead to the shorterT1ρ.We are able to see two types of carbon symmetries in these

minerals based on the values of η. Asymmetry values close to0.5 are structures that resemble the bicarbonate-like carbonenvironment, while values close to 0.1 are structures close tothe D3h magnesite carbon environment (both seen in Scheme1). Static NMR CSA powder patterns are therefore seen to behelpful in determining local carbon structure.Dypingite and hydromagnesite have very similar structures

seen in the chemical shift and asymmetry. One key differ-entiating factor is the relative fraction of carbons in each type ofstructure. Dypingite has a 3:2 ratio of carbons, with three of thecarbon sites having an η value close to 0.1 and two carbon siteswith an η value close to 0.5. However, hydromagnesite has a 1:1ratio of these two sites. Another difference is seen in the T1relaxation times: the hydromagnesite T1 relaxation is longer byapproximately 50 s. By measuring the T1 relaxation time andthe area of the two peaks, dypingite and hydromagnesite areable to be distinguished. These differences will aid inquantifying a mixture of these products since a mixture willexhibit a weighted average of these values. Knowing therelaxation times and integrated areas for each mineral speciesallows quantification of dypingite and hydromagnesite individ-ually from a mixture.The information on each of these minerals is relevant to

future NMR studies, both in situ and ex situ. The in situ NMR

technique gives the full complement of the reacting systemwhile ex situ analysis has the ability to characterize subsets ofthe reaction mixture.In situ NMR measurements are of significant interest because

these reactions can be completed in a reactor that mimics CO2sequestration processes.2 In situ NMR provides a spectroscopic“window” into the product mineral formation, which isimportant because the pH, reactant minerals, and temperatureall affect the product outcome. Such techniques have been usedto observe the speciation of 13C as it is introduced fromenriched 13CO2 gas under geologic sequestration conditions(i.e., ∼80 °C, ∼80 atm).2 These experiments benefit frommethods that allow a high pressure of CO2 to remain on thereacting system which forces the CO2 to remain in solution andthus accessible for reaction. This high pressure reactionenvironment is easily achievable with static 13C NMR, thereforethis technique can monitor the reaction continuously.Ex situ NMR measurements involve collecting the reacted

mineral sample and running 13C{1H} MAS (or CPMAS) NMRexperiments on the product. The reaction products are thendistinguished by their crystallographically inequivalent 13C sites,which are each indicated by an individual 13C chemical shift.This technique can therefore be used to characterize spatialproduct gradients that form during the sequestration reactions,by sectioning the product as a function of depth in a fixed bedreactor or sample chamber. When in situ and ex situ 13C NMRtechniques are used together, the combined information can beintegrated to give a full picture of the reaction and its associatedproducts.

■ ASSOCIATED CONTENT*S Supporting InformationIncluded are Raman spectra, powder X-ray diffraction (pXRD)patterns, 13C{1H} MAS at slow spinning speeds, and 23Na{1H}magic angle spinning (MAS) NMR spectra that confirm theNMR assignments in this article. This material is available freeof charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*Phone: (314) 935-4624; fax: (314) 935-4481; e-mail: hayes@wustl.edu.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe thank the Washington University in St. Louis, Consortiumfor Clean Coal Utilization (CCCU) for funding.

■ REFERENCES(1) Hanchen, M.; Prigiobbe, V.; Baciocchi, R.; Mazzotti, M.Precipitation in the Mg-carbonate system effects of temperatureand CO2 pressure. Chem. Eng. Sci. 2008, 63, 1012−1028.(2) Surface, J. A.; Skemer, P.; Hayes, S. E.; Conradi, M. S. In situmeasurement of magnesium carbonate formation from CO2 usingstatic high-pressure and -temperature 13C NMR. Environ. Sci. Technol.2013, 47, 119−125.(3) Benezeth, P.; Saldi, G. D.; Dandurand, J.-L.; Schott, J.Experimental determination of the solubility product of magnesite at50 to 200°C. Chem. Geol. 2011, 286, 21−31.(4) Zhao, L.; Sang, L.; Chen, J.; Ji, J.; Teng, H. H. Aqueouscarbonation of natural brucite: Relevance to CO2 sequestration.Environ. Sci. Technol. 2010, 44, 406−411.

Environmental Science & Technology Article

dx.doi.org/10.1021/es503390d | Environ. Sci. Technol. XXXX, XXX, XXX−XXXF

(5) King, H. E.; Plumper, O.; Putnis, A. Effect of secondary phaseformation on the carbonation of olivine. Environ. Sci. Technol. 2010,44, 6503−6509.(6) Case, D. H.; Wang, F.; Giammar, D. E. Precipitation ofmagnesium carbonates as a function of temperature, solutioncomposition, and presence of a silicate mineral substrate. Environ.Eng. Sci. 2011, 28, 881−889.(7) Ballirano, P.; De Vito, C.; Mignardi, S.; Ferrini, V. Phasetransitions in the Mg-CO2-H2O system and the thermal decom-position of dypingite, Mg5(CO3)4(OH)2·5H2O: Implications forgeosequestration of carbon dioxide. Chem. Geol. 2013, 340, 59−67.(8) Botha, A.; Strydom, C. A. A. Preparation of magnesium hydroxycarbonate from magnesium hydroxide. Hydrometallurgy 2001, 62,174−183.(9) Lackner, K. S. Carbonate chemistry for sequestering fossil carbon.Annu. Rev. Energy Environ. 2002, 27, 193−232.(10) Wilson, S. A.; Barker, S. L. L.; Dipple, G. M.; Atudorei, V.Isotopic disequilibrium during uptake of atmospheric CO2 into mineprocess waters: Implications for CO2 sequestration. Environ. Sci.Technol. 2010, 44, 9522−9529.(11) Hovelmann, J.; Putnis, C. V.; Ruiz-Agudo, E.; Austrheim, H.Direct nanoscale observations of CO2 sequestration during brucite[Mg(OH)2] dissolution. Environ. Sci. Technol. 2012, 46, 5253−5260.(12) Schaef, H. T.; Windisch, C. F.; McGrail, B. P.; Martin, P. F.;Rosso, K. M. Brucite [Mg(OH2)] carbonation in wet supercriticalCO2: An in situ high pressure X-ray diffraction study. Geochim.Cosmochim. Acta 2011, 75, 7458−7471.(13) Lackner, K. S.; Wendt, C. H.; Butt, D. P.; Joyce, E. L.; Sharp, D.H. Carbon dioxide disposal in carbonate minerals. Energy 1995, 20,1153−1170.(14) Nebel, H.; Neumann, M.; Mayer, C.; Epple, M. On the structureof amorphous calcium carbonate–a detailed study by solid-state NMRspectroscopy. Inorg. Chem. 2008, 47, 7874−7879.(15) Papenguth, H. W.; Kirkpatrick, R. J.; Montez, B.; Sandberg, P. A.

13C MAS NMR spectroscopy of inorganic and biogenic carbonates.Am. Mineral. 1989, 74, 1152−1158.(16) Mason, H. E.; Kozlowski, A.; Phillips, B. L. Solid-state NMRstudy of the role of H and Na in AB-type carbonate hydroxylapatite.Chem. Mater. 2008, 20, 294−302.(17) Sevelsted, T. F.; Herfort, D.; Skibsted, J. 13C chemical shiftanisotropies for carbonate ions in cement minerals and the use of 13C,27Al and 29Si MAS NMR in studies of Portland cement includinglimestone additions. Cem. Concr. Res. 2013, 52, 100−111.(18) Hoyt, D. W.; Turcu, R. V. F.; Sears, J. A.; Rosso, K. M.; Burton,D.; Felmy, A. R.; Hu, J. Z.; Burton, S. D. High-pressure magic anglespinning nuclear magnetic resonance. J. Magn. Reson. 2011, 212, 378−385.(19) Kwak, J. H.; Hu, J. Z.; Hoyt, D. W.; Sears, J. A.; Wang, C.;Rosso, K. M.; Felmy, A. R. Metal carbonation of forsterite insupercritical CO2 and H2O using solid state 29Si, 13C NMRspectroscopy. J. Phys. Chem.C 2010, 114, 4126−4134.(20) Feng, J.; Lee, Y. J.; Reeder, R. J.; Phillips, B. L. Observation ofbicarbonate in calcite by NMR spectroscopy. Am. Mineral. 2006, 91,957−960.(21) Bryce, D. L.; Bultz, E. B.; Aebi, D. Calcium-43 chemical shifttensors as probes of calcium binding environments. Insight into thestructure of the vaterite CaCO3 polymorph by 43Ca solid-state NMRspectroscopy. J. Am. Chem. Soc. 2008, 130, 9282−9292.(22) Pines, A.; Rhim, W. K.; Waugh, J. S. 13C chemical shieldinganisotropy in solids. CS2 and CaCO3. J. Chem. Phys. 1971, 54, 5438−5440.(23) Lauterbur, P. Anisotropy of the C13 chemical shift in calcite.Phys. Rev. Lett. 1958, 1, 343−344.(24) Sandengen, K.; Jøsang, L. O.; Kaasa, B. Simple method forsynthesis of magnesite (MgCO3). Ind. Eng. Chem. Res. 2008, 47,1002−1004.(25) Sayles, F. L.; Fyfe, W. S. The crystallization of magnesite fromaqueous solution. Geochim. Cosmochim. Acta 1973, 37, 87−99.

(26) Felmy, A. R.; Qafoku, O.; Arey, B. W.; Hu, J. Z.; Hu, M.; ToddSchaef, H.; Ilton, E. S.; Hess, N. J.; Pearce, C. I.; Feng, J.; et al.Reaction of water-saturated supercritical CO2 with forsterite: Evidencefor magnesite formation at low temperatures. Geochim. Cosmochim.Acta 2012, 91, 271−282.(27) Schaef, H. T.; McGrail, B. P.; Loring, J. L.; Bowden, M. E.; Arey,B. W.; Rosso, K. M. Forsterite [Mg2SiO4] carbonation in wetsupercritical CO2: An in situ high-pressure X-ray diffraction study.Environ. Sci. Technol. 2013, 47, 174−181.(28) Davies, P. J.; Bubela, B. The transformation of nesquehoniteinto hydromagnesite. Chem. Geol. 1973, 12, 289−300.(29) Surface, J. A. In situ high pressure and temperature 13C NMRfor the study of carbonation reactions of CO2. Ph.D. Dissertation,Washington University in St. Louis, St. Louis, MO, 2013.(30) Kloprogge, J. T.; Martens, W. N.; Nothdurft, L.; Duong, L. V.;Webb, G. E. Low temperature synthesis and characterization ofnesquehonite. J. Mater. Sci. Lett. 2003, 22, 825−829.(31) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calve, S.;Alonso, B.; Durand, J.-O.; Bujoli, B.; Gan, Z.; Hoatson, G. Modellingone- and two-dimensional solid-state NMR spectra. Magn. Reson.Chem. 2002, 40, 70−76.(32) Giammar, D. E.; Bruant, R. G.; Peters, C. A. Forsteritedissolution and magnesite precipitation at conditions relevant for deepsaline aquifer storage and sequestration of carbon dioxide. Chem. Geol.2005, 217, 257−276.(33) Xiong, W.; Giammar, D. Forsterite carbonation in zones withtransport limited by diffusion. Environ. Sci. Technol. Lett. 2014, 1, 333−338.(34) Akao, M.; Marumo, F.; Iwai, S. The crystal structure ofhydromagnesite. Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst.Chem. 1974, 30, 2670−2672.(35) Murdoch, J. Unit cell of hydromagnesite. Am. Mineral. 1954, 39,24−29.(36) Akao, M.; Iwai, S. The hydrogen bonding of hydromagnesite.Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem. 1977, 33,1273−1275.(37) Canterford, J. H.; Tsambourakis, G.; Lambert, B. Someobservations on the properties of dypingite, Mg5(CO3)4(OH)2·5H2O, and related minerals. Mineral. Mag. 1984, 48, 437−442.(38) Raade, G. Dypingite, a new hydrous basic carbonate ofmagnesium, from Norway. Am. Mineral. 1970, 55, 1457−1465.(39) Yi, Y.; Liska, M.; Unluer, C.; Al-Tabbaa, A. Carbonatingmagnesia for soil stabilisation. Can. Geotech. J. 2013, 50, 899−905.(40) Power, I. M.; Wilson, S. A.; Thom, J. M.; Dipple, G. M.;Southam, G. Biologically induced mineralization of dypingite bycyanobacteria from an alkaline wetland near Atlin, British Columbia,Canada. Geochem. Trans. 2007, 8.(41) Hales, M. C.; Frost, R. L.; Martens, W. N. Thermo-ramanspectroscopy of synthetic nesquehoniteImplication for the geo-sequestration of greenhouse gases. J. Raman Spectrosc. 2008, 39, 1141−1149.(42) Dong, M.; Li, Z.; Mi, J.; Demopoulos, G. P. Solubility andstability of nesquehonite (MgCO3·3H2O) in mixed NaCl + MgCl2,NH4Cl + MgCl2, LiCl, and LiCl + MgCl2 solutions. J. Chem. Eng. Data2009, 54, 3002−3007.(43) Stephan, G. W.; MacGillavry, C. H. The crystal structure ofnesquehonite, MgCO3.3H2O. Acta Crystallogr., Sect. B: Struct.Crystallogr. Cryst. Chem. 1972, 28, 1031−1033.(44) Ferrini, V.; De Vito, C.; Mignardi, S. Synthesis of nesquehoniteby reaction of gaseous CO2 with Mg chloride solution: Its potentialrole in the sequestration of carbon dioxide. J. Hazard. Mater. 2009,168, 832−837.(45) Konigsberger, E.; Konigsberger, L.-C.; Gamsjager, H. Low-temperature thermodynamic model for the system Na2CO3−MgCO3−CaCO3−H2O. Geochim. Cosmochim. Acta 1999, 63, 3105−3119.(46) Giester, G.; Lengauer, C. L.; Rieck, B. The crystal structure ofnesquehonite, MgCO3·3H2O, from Lavrion, Greece. Mineral. Petrol.2000, 70, 153−163.

Environmental Science & Technology Article

dx.doi.org/10.1021/es503390d | Environ. Sci. Technol. XXXX, XXX, XXX−XXXG

(47) Kwak, J. H.; Hu, J. Z.; Turcu, R. V. F.; Rosso, K. M.; Ilton, E. S.;Wang, C.; Sears, J. a.; Engelhard, M. H.; Felmy, A. R.; Hoyt, D. W.The role of H2O in the carbonation of forsterite in supercritical CO2.Int. J. Greenhouse Gas Control 2011, 5, 1081−1092.(48) Fukushima, E.; Roeder, S. B. W. Experimental Pulse NMR: ANuts and Bolts Approach; Addison-Wesley Publishing Company, 1981.

Environmental Science & Technology Article

dx.doi.org/10.1021/es503390d | Environ. Sci. Technol. XXXX, XXX, XXX−XXXH